Different output when parsing HTML

[longcdf]

Hi,
I'm facing some strange behavior when trying to parse a HTML file using LibTidy.
I have a html file contains something likes this:

<li class="ikonpunkt span8"><a href="http://www.vegvesen.no/jobb/Ledige+stillinger" class="menuitem"><img alt="Ledige stillinger hvit" src="http://www.vegvesen.no/jobb/For+studenter+og+nyutdannede/nyheter/_attachment/872915/binary/1031072?_ts=14d1da193f0"/><h3>Ledige stillinger</h3></a></li>

I'm using LibTidy to parse this file.
And it randomly generate 2 outputs:
1 -

<li class="ikonpunkt span8"><a href="http://www.vegvesen.no/om+statens+vegvesen/kontakt+oss" class="menuitem"><img alt="Kontakt oss hvit" src="http://www.vegvesen.no/jobb/For+studenter+og+nyutdannede/nyheter/_attachment/873039/binary/1031133?_ts=14d1dd66378">
<h3>Kontakt oss</h3>
</a></li>

2-

<li class="ikonpunkt span8"><a href="http://www.vegvesen.no/om+statens+vegvesen/kontakt+oss" class="menuitem"><img alt="Kontakt oss hvit" src="http://www.vegvesen.no/jobb/For+studenter+og+nyutdannede/nyheter/_attachment/873039/binary/1031133?_ts=14d1dd66378"></a>
<h3><a href="http://www.vegvesen.no/om+statens+vegvesen/kontakt+oss" class="menuitem">Kontakt oss</a></h3>
</li>

It happens very randomly.
Could you please help me explain it and suggest a way to overcome this issue?
Thanks

[geoffmcl]

@longcdf thank you for the issue... but at this time do not understand...

Using your input, and a config of -w 0, on current 5.7.17, I repeat the output of 1., given an obvious different name <h3>Ledige stillinger</h3>... but that error aside...

What version of libTidy? Hopefully https://github.com/htacg/tidy-html5 next source...

How are you using libTidy? In what container, app, lib, whatever... src...

A random happening can only be explained as due to configuration, at the moment libTidy runs... care with multiple threads...

A library, like libtidy.a, has to produce the same output, given a config, and an input, a gzillion times over... forever... no random change... not possible... code paths can not change, without change...

So, at this moment, I really do not see how libTidy could very randomly output 2...

Even the sense is changed - the <h3> header is also a link to the URL... another anchor <a ...>name</a> added... the input must have changed...

At this time do not understand the problem... more information needed... thanks...

[longcdf]

Hi @geoffmcl , thank you for your response.
I'm using tidys.lib 5.6.0 downloaded from http://binaries.html-tidy.org/
This issue happens when I'm using multi-threading.
I'm having about 20 threads, each thread will parse a different html file.
I tested by using tidyParseFile followed by a tidySaveFile to make sure I didn't touch the tinyDoc.

Original file: https://drive.google.com/open?id=1Ch8QarwXZfpz_KUBSdoQk2cORWBItmGb
Correct output file: https://drive.google.com/open?id=1DxW27av3umJYZjQhFBycQPvgpXm_WGNi
Randomly wrong output file: https://drive.google.com/open?id=1pwEJMpPlhEG3vZ5g6uS0T_NihV5b3wT4

One more note is it's not always failed exactly like the second output file. Maybe less or more differences but always has problem at the <h3> tag.

So I wonder maybe the multithreading cause the issue? Cause when I'm using single thread it is fine.

Thanks.

[balthisar]

Looks like this has been address. Please feel to reopen if I'm wrong.